A Systematic Review of Deep Learning Applications in Breast Cancer Detection Using MRI, Mammography, and Ultrasound

A Systematic Review of Deep Learning Applications in Breast Cancer Detection Using MRI, Mammography, and Ultrasound

Asma Raisi1 Mahsa Nasiri2 Hajar Danesh3

1) M.Sc. Student of Biomedical Engineering
2) B.Sc. in Biomedical Engineering
3) Ph.D. in Biomedical Engineering

IOI Code : ACCD-ACZDAH
Publication : 10th International Conference on Applied Researches in Science & Engineering (carseconf.com)
Abstract :
    Breast cancer remains one of the most prevalent and lethal malignancies among women worldwide, where early detection plays a pivotal role in improving patient survival rates. In recent years, the integration of medical imaging modalities with artificial intelligence algorithms—particularly convolutional neural networks (CNNs)—has substantially enhanced diagnostic accuracy and speed. In this study, a systematic search was conducted across reputable databases. Articles were screened based on predefined eligibility criteria, and qualified studies were systematically analyzed and compared. The reported findings demonstrate that CNN-based models achieve automated classification of breast lesions with an accuracy ranging from 90% to 98%, sensitivity up to 96%, and an Area Under the Curve (AUC) reaching 0.98. Notably, these models superiorly outperformed traditional methods in identifying small masses within dense tissues via MRI and in reducing operator dependency in ultrasound image interpretation. Furthermore, deploying deep learning-based systems yielded reduced screening times and minimized human error, particularly in borderline cases. Overall, the evidence suggests that fusing multimodal imaging data with tailored deep learning architectures can serve as a robust clinical decision-support tool, significantly improving breast cancer diagnostic precision in real-world settings.

Keywords : breast cancer deep learning convolutional neural network mammography mri ultrasound